Issue computing NGAEast GMMs with vectorized input about oq-engine

[micheles]

Using a GSIM is non-obvious. First of all, it should not be instantiated directly, but through the valid.gsim function. Secondly, contexts objects should be instantiated by using a ContextMaker instance, which has to be instantiated carefully. Thirdly, the .compute method should not be called directly. In short, here is how to do what you want:

import numpy as np
from openquake.hazardlib import valid
from openquake.hazardlib.contexts import simple_cmaker

gsim = valid.gsim('''[NGAEastGMPE]
gmpe_table="NGAEast_PEER_EX.hdf5"''')
n = 10
mags = np.linspace(5, 7, n)
cmaker = simple_cmaker([gsim], ['PGA'], mags=['%.2f' % mag for mag in mags])
ctx = cmaker.new_ctx(n)
ctx["mag"] = mags
ctx["vs30"] = np.linspace(200, 300, n)
ctx["rrup"] = np.linspace(1, 100, n)

mean, sigma, tau, phi = cmaker.get_mean_stds([ctx])

[smithUSGS]

Thank you, I adapted your example and it passed the test. Note that I didn't need to instantiate the GMM using valid.gsim (although I can and it is what you recommend). If you don't mind explaining why using valid.gsim is a good practice, I would appreciate that. Either way, I think this issue can be closed now.

[micheles]

Using valid.gsim is a good practice because it is the official way the engine convert strings into GMPEs, so it is guaranteed to work forever, while manual instantiation changed in the past and could potentially change again in the future.

[emthompson-usgs]

Thanks @micheles, we weren't aware of this. Good to know.